IRC channel logs

2021-09-06.log

back to list of logs

<zimoun>hi!
<civodul>hey zimoun!
<civodul>how's everything?
<zimoun>fine and you. :-) Well, I have a short presentation of Guix in a local “bioinfo” seminar. Hard to find the correct level… ;-)
<civodul>oh neat
<civodul>i'm sure you'll find the right words :-)
<zimoun>and because I also have the JRES tuto, I am testing some of my speech. ;-)
<civodul>heh
<civodul>i was thinking we could have a call among interested parties to discuss what we'd like to work on in an HPC context in the coming months
<zimoun>yes, let do that! :-) I will try to drop an email if no one beats me.
<civodul>awesome, thanks!
<civodul>we can email guix-science i guess
<civodul>(we'd also need to do a similar thing Guix-wide, i suppose...)
<zimoun>yes, we could do something as a virtual Guix Day
<rekado>I’d very much like to resume work on the GWL, but I’ve been feeling very exhausted. If any of you have any use cases for the GWL I’d love to hear them to get motivated to hack on it.
<civodul>oh yes, that'd be great!
<civodul>so we need to bring you energy in the form of feature requests? :-)
<rekado>correct :)
<rekado>I have been collecting a few annoyances with snakemake, but I don’t know if it’s feasible for me to improve on those things.
<rekado>there are a few annoyances with GWL, too (mostly due to Guix performance problems), but I think I can fix those without too much work.
<rekado>one thing that turned out to be problematic is how strongly workflow languages are tied to files.
<rekado>file-based inputs and outputs makes things easy, but sometimes you just want to do without serialization and deserialization and pass memory blocks around
<rekado>but the biggest issue I have is lack of users. PiGx doesn’t use the GWL, which means that any work on the GWL is on top of the work I already do. Don’t really have enough time for that at the moment.
<zimoun>rekado: about GWL, I agree about file-centric. My mind is unclear so I am trying to read in deep <https://www.microsoft.com/en-us/research/uploads/prod/2020/04/build-systems-jfp.pdf> to get a better image of the picture.
<zimoun>in my lab, people are not enough maverick to use GWL )-: And they are already fighting the snake. :-)
<civodul>oh that's a followup to the original paper?